Abstract:

I present the Central-peripheral Dichotomy (CPD) theory (Zhaoping 2017, 2019) together with visual psychophysics tests of its predictions. CPD states that central vision is specialized for seeing (recognizing), and peripheral vision for looking (shifting gaze/attention). Given an information bottleneck that is proposed to start with the connection from the primary visual cortex (V1) to higher visual areas, CPD suggests that mainly central rather than peripheral vision enjoys rich top-down feedback that aids seeing. Two theoretical predictions follow: First, some visual illusions should be visible in peripheral, but not central, vision, because peripheral vision is misled by the impoverished feed-forward information that makes it through the bottleneck from V1. Central vision avoids this fate by exploiting top-down feedback. Second, central vision should be susceptible to the same illusions when top-down feedback is compromised, for instance by backwards masking. Both predictions are confirmed in a reversed depth illusion created using random-dot stereograms. These behavioral confirmations also suggest an easily feasible paradigm for neurophysiological studies in behaving animals on how the recurrent processing for visual recognition could be implemented, to further test the CPD theory.